Integrating time and space in tropical phenology to track climate change

Monitor and track plant responses to environmental cues across the tropics remains a challenge and a key issue to the global warming research. Traditional on-the-ground direct, manned phenological observations are laborious and time consuming, restricted to local areas and long time-intervals, usually monthly. Near-surface remote phenology using repeated photographs taken from digital cameras or phenocams, usually set up at the top of towers have reduced the human labor constraints, increased the frequency of observation (hour to daily), and eliminates the uncertainty of cloud cover, enhancing the resolution of information at individual tree, species, and community scales. Furthermore, phenocams have proven to be an effective tool for simultaneously monitoring from several species to ecosystems, accurately accessing leaf changes daily and relate to climate drivers. However, phenocams are still spatially limited and the use of drones has helped to increase spatial cover and resolution but again at the expenses of a short temporal frequency (bi-weekly to monthly). Here we discuss how to combine new technologies to traditional phenology, integrating at time and space and enhancing the capabilities of phenological observations to detect changes on vegetation phenology at various scales, from leaves to ecosystems. Our studies have been carried out in distinct Brazilian ecosystem, from dry forests to rainforests, cerrado and grasslands, all holding high species richness and endemism. We argue that the combination of technologies framed within a e-science research project and using machine learning and other computational tools has improved our observational capabilities. We are better integrating time and space, increasing our accuracy to relate phenology to environmental cues and, as a consequence, predict potential responses to climate changes. (Financial support: FAPESP #201350155-0; CNPq; CAPES)